改进的货运量最优变权组合预测模型

An improved combined model based on optimal variable weight for prediction of freight volumes

提出一种改进的变权组合预测模型,实现货运量的短期和长期预测。该方法采用多元线性回归、二次指数平滑法、灰色模型、指数曲线外推方法分别对货运量进行预测,设计一个多项式的系数矩阵,建立组合预测模型。为了保证组合模型长期预测的稳定性,在进行长期预测时,在模型中增加了收敛条件。最后,通过引入松弛函数,避免求解方程时的病态,利用拉格朗日极值法求解该组合预测模型的系数矩阵,获得组合预测模型。根据长沙市1990年~2008年货运量的原始数据,采用该模型进行短期和长期预测,结果表明了该模型的有效性。

An improved combined model based on variable weight was proposed to realize short term and long term prediction of freight volumes.Freight volumes were first predicted through multiple linear regression,quadratic exponential smoothing,gray model and exponential curve extrapolation method respectively,and then coefficient matrix was designed.So we can obtain the combined model.A convergence condition was integrated into the model in order to ensure the stability of long-term forecast.At last,relaxation function was introduced to avoid the sick of solving equations,Lagrange operator was employed to solve the combined model to obtain the coefficient matrix,and so the combined forecasting model can be solved.According to Changsha from 1 990 to 2 008,freight volumes of original data are predicted through the model for short-term and long-term,and the results show the the model is valid.